Sparse Decoding (MHA) on Synthetic 128K Context (H100, FP16)
0.7FlashInfer Latency (ms)
FlashInfer
Updated 8d ago
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
2026.05 | 0.7 | - | - | - | - | - | - | |
2026.05 | 2.79 | - | - | - | - | - | - | |
2026.05 | 5.61 | - | - | - | - | - | - | |
2026.05 | 11.18 | - | - | - | - | - | - | |
| - | 0.91 | 1.62 | 2.18 | 2.68 | 3.14 | 3.37 | ||
| - | 1.02 | 1.86 | 2.56 | 3.17 | 3.74 | 4 | ||
| - | 1.12 | 2 | 2.71 | 3.32 | 3.86 | 4.11 | ||
| - | 1.22 | 2.13 | 2.84 | 3.43 | 3.94 | 4.17 |